optimize tri further

This commit is contained in:
bendn 2023-10-05 08:13:30 +07:00
parent 853b51fc50
commit 1c218cd1a9
No known key found for this signature in database
GPG key ID: 0D9D3A2A3B2A93D6
2 changed files with 10 additions and 15 deletions

View file

@ -1,9 +1,7 @@
//! trongle drawing //! trongle drawing
use crate::math::madd; use crate::math::madd;
use std::cmp::{max, min};
use crate::Image; use crate::Image;
use std::cmp::{max, min};
impl<T: AsMut<[u8]> + AsRef<[u8]>, const CHANNELS: usize> Image<T, CHANNELS> { impl<T: AsMut<[u8]> + AsRef<[u8]>, const CHANNELS: usize> Image<T, CHANNELS> {
/// Draw a (filled) triangle /// Draw a (filled) triangle
@ -17,31 +15,26 @@ impl<T: AsMut<[u8]> + AsRef<[u8]>, const CHANNELS: usize> Image<T, CHANNELS> {
/// (1.0, 8.0), // point c /// (1.0, 8.0), // point c
/// [255] // white /// [255] // white
/// ); /// );
/// # assert_eq!(a.buffer(), b"\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\xff\x00\x00\x00\x00\x00\x00\x00\xff\xff\xff\xff\x00\x00\x00\x00\x00\xff\xff\xff\xff\xff\xff\x00\x00\x00\x00\xff\xff\xff\xff\xff\xff\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00"); /// # assert_eq!(a.buffer(), b"\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\xff\x00\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\x00\x00\x00\x00\x00\x00\x00\x00\xff\xff\xff\x00\x00\x00\x00\x00\x00\xff\xff\xff\xff\xff\x00\x00\x00\x00\x00\xff\xff\xff\xff\xff\xff\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00");
/// ``` /// ```
pub fn tri( pub fn tri(
&mut self, &mut self,
(x1, y1): (f32, f32),
(x2, y2): (f32, f32), (x2, y2): (f32, f32),
(x1, y1): (f32, f32),
(x3, y3): (f32, f32), (x3, y3): (f32, f32),
c: [u8; CHANNELS], c: [u8; CHANNELS],
) { ) {
// TODO optimize
let ymin = max(y1.min(y2).min(y3) as u32, 0); let ymin = max(y1.min(y2).min(y3) as u32, 0);
let ymax = min(y1.max(y2).max(y3) as u32, self.height()); let ymax = min(y1.max(y2).max(y3) as u32, self.height());
let xmin = max(x1.min(x2).min(x3) as u32, 0); let xmin = max(x1.min(x2).min(x3) as u32, 0);
let xmax = min(x1.max(x2).max(x3) as u32, self.width()); let xmax = min(x1.max(x2).max(x3) as u32, self.width());
for y in ymin..ymax { for y in ymin..ymax {
for x in xmin..xmax { for x in xmin..xmax {
let s = madd(x1 - x3, y as f32 - y3, -(y1 - y2) * (x as f32 - x3)); // algorithm from https://web.archive.org/web/20050408192410/http://sw-shader.sourceforge.net/rasterizer.html, but im too dumb to implement the faster ones
let t = madd(x2 - x1, y as f32 - y1, -(y2 - y1) * (x as f32 - x1)); if madd(x1 - x2, y as f32 - y1, -(y1 - y2) * (x as f32 - x1)) > 0.
&& madd(x2 - x3, y as f32 - y2, -(y2 - y3) * (x as f32 - x2)) > 0.
if (s < 0.0) != (t < 0.0) && s != 0.0 && t != 0.0 { && madd(x3 - x1, y as f32 - y3, -(y3 - y1) * (x as f32 - x3)) > 0.
continue; {
}
let d = madd(x3 - x2, y as f32 - y2, -(y3 - y2) * (x as f32 - x2));
if d == 0.0 || (d < 0.0) == (s + t <= 0.0) {
// SAFETY: x, y are bounded // SAFETY: x, y are bounded
unsafe { self.set_pixel(x, y, c) }; unsafe { self.set_pixel(x, y, c) };
} }

View file

@ -1,4 +1,6 @@
//! utility math
/// Calculates `a * b + c`, with hardware support if possible. /// Calculates `a * b + c`, with hardware support if possible.
#[allow(clippy::suboptimal_flops)]
pub fn madd(a: f32, b: f32, c: f32) -> f32 { pub fn madd(a: f32, b: f32, c: f32) -> f32 {
if cfg!(target_feature = "fma") { if cfg!(target_feature = "fma") {
a.mul_add(b, c) a.mul_add(b, c)